Research Article
Drug-Drug Interactions Prediction Using Fingerprint Only
Table 5
Comparison of fingerprint- and network-based classifiers under two types of tenfold cross-validation.
| | Cross-validation | Classifier | Model | Accuracy | Precision | Recall | F1-measure | MCCa |
| | Entire tenfold cross-validation | Fingerprint-based classifier (random forest) | Addition + subtraction | 86.86% | 89.21% | 85.23% | 87.16% | 73.82% | | Network-based classifier (random forest) | 77.88% | 77.60% | 78.06% | 77.81% | 55.79% | | Composition tenfold cross-validation | ODITb test dataset | Fingerprint-based classifier (random forest) | Addition + Hadamard | 74.31% | 72.43% | 75.46% | 73.77% | 48.83% | | Network-based classifier (random forest) | 70.79% | 66.72% | 72.65% | 69.31% | 41.94% | | NDITc test dataset | Fingerprint-based classifier (random forest) | Addition + Hadamard | 59.99% | 33.23% | 70.80% | 43.48% | 23.34% | | Network-based classifier (random forest) | 57.53% | 41.20% | 62.17% | 48.49% | 16.42% |
|
|
aMathews Correlation Coefficient. bODIT: One Drug In Train set. cNDIT: No Drug In Train set.
|
